Our Work
We aim to streamline health AI evaluation and governance, building robust frameworks and methodologies for responsible use and oversight while raising awareness of best practices within the community.
The Duke Health AI Evaluation & Governance (E&G) Program advances the responsible integration of artificial intelligence (AI) into healthcare. Through strategic initiatives, collaborative partnerships, and thought leadership, we ensure AI solutions are safe, effective, and contribute to reducing disparities while delivering high-quality care.
Building Foundations for Responsible AI
Algorithm-Based Clinical Decision Support (ABCDS) Oversight Initiative
Our flagship ABCDS Oversight initiative exemplifies Duke's commitment to clinical excellence and patient safety. This comprehensive program establishes governance, evaluation, and monitoring standards for AI-driven solutions intended for use across the Duke Health. By adhering to guiding principles such as clinical value, fairness, usability, regulatory compliance, and accountability, the initiative ensures AI tools are rigorously assessed at every stage of their lifecycle — from development to deployment and ongoing monitoring. Through the ABCDS Oversight Initiative, we set a benchmark for responsible AI use in healthcare, promoting transparency and trustworthiness across the system.
Connecting Experts Across Disciplines
Duke-Margolis Institute for Health Policy Collaboration
The Duke Health AI E&G Program and the Duke-Margolis Institute for Health Policy share a strong partnership grounded in a mutual commitment to advancing responsible and patient-centered healthcare innovation. Together, we aim to align AI development with regulatory standards and clinical priorities, reinforcing Duke Health’s vision of leading transformative change in healthcare. As part of this collaboration, we co-authored a White Paper titled "AI Governance in Health Systems: Aligning Innovation, Accountability, and Trust." This important work addresses critical aspects of AI governance in healthcare, providing actionable recommendations for fostering innovation while ensuring accountability and trust. To further this effort, Duke-Margolis hosted a webinar bringing together experts from healthcare, policy, and technology to explore practical strategies for implementing the White Paper’s recommendations.
Scaling for National & International Impact
Partnerships
Coalition for Health AI (CHAI): Duke Health is a founding member of the Coalition for Health AI (CHAI), a non-profit organization dedicated to advancing best practices for responsible AI in healthcare. CHAI convenes a diverse network of stakeholders, including health systems, researchers, technology developers, patient advocacy groups, and community-based providers, to develop consensus-driven guidelines for trustworthy AI. The Duke Health AI Evaluation & Governance team was instrumental in CHAI’s formation and its evolution from a volunteer-led initiative to an established non-profit entity. Duke Health remains an active contributor, providing expertise and supporting CHAI’s mission to promote the responsible adoption of AI in healthcare.
Trustworthy & Responsible AI Network (TRAIN)TM: As a founding member of the Trustworthy & Responsible AI Network (TRAIN), Duke Health is dedicated to operationalizing and democratizing responsible AI in healthcare. TRAIN serves as a collaborative hub where member organizations share insights on procuring, adopting, and implementing AI solutions. It provides healthcare institutions with the tools and frameworks needed to integrate AI ethically and effectively. Duke Health plays a critical role in advancing TRAIN’s mission by contributing expertise in AI validation, safety, and monitoring. Through this collaboration, Duke helps drive the responsible and trustworthy deployment of AI across diverse healthcare settings, ensuring that AI technologies enhance patient outcomes and care while addressing issues of fairness, transparency, and accountability.
Collaborations
Health AI Maturity Model Project: Duke Health and Vanderbilt University Medical Center (VUMC) are leading the development of the Health AI Maturity Model, a structured framework designed to help health systems evaluate their readiness for AI adoption. Supported by a $1.25 million grant from the Gordon and Betty Moore Foundation, this initiative focuses on critical areas including governance, data quality, workforce competencies, and ongoing monitoring. The resulting framework will serve as a nationwide blueprint, enabling health systems to benchmark their AI capabilities and implement best practices for responsible AI governance.
Through these initiatives and partnerships, the Duke Health AI E&G Program is shaping a future where AI-powered healthcare enhances outcomes while prioritizing safety, fairness, and accountability.
